A persistent autonomous engineering harness runtime that transforms any repository into a self-improving software system. Use this skill whenever the user wants to: build or run an autonomous coding agent, set up a self-healing engineering loop, orchestrate multi-agent software development, implement harness engineering principles, create a doc-driven development workflow, or run long-horizon autonomous software tasks. Trigger on phrases like "autonomous agent", "self-healing codebase", "harness engineering", "multi-agent pipeline", "continuous improvement loop", "OpenClaw skill", or any request for an agentic engineering system that runs without constant human input. Primarily designed for Claude Code and OpenClaw environments.